Research And Realization Of Robot Intelligent Demonstration Learning System In Chemical Analysis

With the development of intelligent technology,the application scenarios of robots have gradually become wider due to their advantages of precision,automation and strong plasticity.In the chemical field,chemical analysis experiments are an essential part of the research,study and production process in this field.However,at present,the level of intelligence in chemical experiments is not high.In order to accelerate scientific and technological innovation and build intelligent laboratories,my country has issued a series of supporting policies.Therefore,giving full play to the advantages of robots and using them to replace laboratory workers to complete some cumbersome and high-risk experiments will become the main direction of the intelligent development of chemical laboratories.In this process,the use of intelligent equipment such as robots and robotic arms requires the mastery of complex programming and control techniques,which is difficult for chemical experimenters.This makes the problem of how to control robots,robotic arms and other intelligent devices to complete chemical experiments in a simpler and faster way.In order to solve this problem,this paper uses a desktop experimental manipulator as a physical carrier,and combines image processing,speech recognition,information fusion and equipment control technologies to propose a robotic arm intelligent teaching solution for chemical experimental scenarios.The plan uses the desktop robotic arm in chemical analysis experiments,and helps the robotic arm to acquire experimental skills through the teaching methods of "doing while talking" by the experimenter and "watching and listening while the robotic arm",so as to improve the intelligence level of the experimental robotic arm.while lowering its control programming threshold.In this regard,this paper has done the following work:（1）Research on video-based teaching algorithms.The target detection algorithm is studied,and the applicability of Open Pose to action detection is tested.Finally,an improved two-stream convolutional network is used to extract the features and behaviors of the experimenter’s arm movements,and obtain the experimental action primitives and confidence probability.The sequential combination is the sequence of experimental action groups,which realizes the teaching based on video vision.（2）Research on the teaching algorithm based on speech.The speech text is recognized by speech recognition technology,and the action keywords in the speech text spoken by the experimenter are extracted by the keyword extraction algorithm of regular expression.（3）Study the teaching algorithm improvement strategy based on audio-visual information fusion to improve the accuracy of skill acquisition during teaching.The information fusion algorithm is studied,and the audio-visual information fusion scheme is proposed according to the physiological enlightenment of the information processing mechanism of the human brain,and the particularity of the scheme used in the robotic arm is studied.Using the FEI-DEO fusion strategy,the action information obtained from the video and audio parts is fused at the decision-making layer,and a high-accuracy experimental action group is output.（4）Research on the reproduction of the experimental action based on the motion control of the manipulator.According to the obtained experimental action sequence,the robotic arm performs kinematics analysis based on the DH method,uses the inverse solution method to control the operation of the stepper motor according to the position of the end effector;uses the five-dimensional representation to describe the grasping pose of the object,and uses the chemical equipment to describe the grasping pose.Particularity proposes a label grabbing scheme to realize grabbing and assist in the reproduction of complete actions.Finally,the overall and single-module operation of the robotic arm teaching system in this paper is tested.The results show that the use of audio-visual information fusion improves the system’s skill acquisition accuracy by about 5%-7%;the robotic arm further performs motion control.When reproducing the action,the overall teaching success rate is about 85.5%,which verifies the effectiveness of the solution of teaching the robotic arm to replace the manual completion of chemical analysis experiments by teaching.